Saxo/Linkbricks-Horizon-AI-Korean-Superb-27B
Full fine-tune vs google/gemma-2-27b-it. Fine-tune: 100% of the weights changed, across 46 of 46 layers.
Tensors changed: Embeddings 1/1, Norms 64/185, MLP 138/138, Attention 184/184
3 derivatives checked
- mradermacher/Linkbricks-Horizon-AI-Korean-Superb-27B-i1-GGUF: Quantization
- mradermacher/Linkbricks-Horizon-AI-Korean-Superb-27B-GGUF: Quantization
- kfawel/pandipirraloda: Full fine-tune (fewer layers)